2026-09-28
Going on a Tangent with the Intel 8087’s Hybrid CORDIC Algorithm
Continuing their reverse-engineering of Intel’s 8087 FPU, [Ken Shirriff] and friends took a look at one of the trigonometric functions, specifically FPTAN. The most exciting part with such reverse-engineering is...
The Register thinks my La Croix benchmark is legit (sorta)
Brandon Vigliarolo: iPhone 18 Pro benchmark boosted by giving it a cold drinkTo make a long story short, a cold can of La Croix (in Birchler’s case – the cooling...
Germany’s Battleship Bismarck Displaced More Than 50,000 Tons. A British Biplane That Flew Under 140 MPH Jammed Its Rudders.
The Fairey Swordfish first flew in 1934 and joined the Fleet Air Arm in 1936, and by 1941 it looked like an antique: a fabric-covered biplane with an open cockpit,...
From Australia? Consider Contacting Your Representative to Express Concern Over OpenAI Attack
Recently, an OpenAI agent allegedly hacked into a website run by the Australian government, apparently in order to access statistical data. While the hack did not seem to compromise any...
The US Navy Brought Back Just One Battleship for Vietnam. USS New Jersey Fired More 16-Inch Shells There Than in World War II and Korea Combined.
By 1968, North Vietnam had grown extremely proficient at defending its airspace against American warplanes, so the Navy turned to a platform Hanoi would find far harder to destroy, and...
Russia’s Yasen-M Class Nuclear Submarine Might Be the Biggest Threat to the US Navy Right Now
In the 1980s, Washington and Moscow both set out to build a new generation of attack submarines. The U.S. planned 29 Seawolf-class boats, built only three, and switched to the...
Riley Hill on the Fold8's crease
Riley Hill: Galaxy Z Fold 8 Review: Worth Buying in YearsAs for the crease, it’s not something you see when looking at the larger display head-on. Tilt the phone slightly...
Sector: de-slopping AI-driven research
Status: as an AI Safety researcher working with LLMs daily over the last 2 years, I think I found some best practices and synthesized them into a single "Sector workflow"...
The Quest for Embedded Evaluators
Dario Amodei’s essay We Must Pace the Frontier committed Anthropic to embedded evaluators, who would be placed inside Anthropic and given employee-level access, so they could provide outside perspective and...
Sukhoi’s Su-37 Terminator Fighter Could Hit Mach 2.5 and Flip a Full 360 Degrees in a Tiny Patch of Sky. Only 1 Was Built and No Buyer Ever Paid to Put It Into Production.
The Su-37 began as T10M-11, the 11th of Sukhoi's Su-27M prototypes. Sukhoi fitted it with Lyulka-Saturn AL-37FU turbofans whose nozzles swiveled about 15 degrees up or down, so the engines...
2026-09-27
Why AI Agents Forget Everything (and How to Build Ones That Don't)
Learn how agent memory captures events, retrieves relevant context, and forgets outdated information to make AI interactions more useful and personal.
Designing Agent Memory for Freshness, Supersession, and Retention
AI agents can retain outdated or sensitive information. The FRESH Memory Model offers a practical way to make agent memory safer and more reliable.
The SDP Offer/Answer Rules That Bite During WebRTC Renegotiation
Why WebRTC renegotiation breaks: a practical guide to SDP m-lines, transceiver ordering, rollback, implicit descriptions, and negotiation errors.
How Much Should AI Be Allowed to Decide in Your Product? A PM's Decision Tree
"Can AI do this?" is the wrong question. A PM's decision tree for sorting AI features by what a wrong answer costs and who absorbs it.
FaveNest
Bookmarks that organize themselves with Apple Intelligence Discussion | Link
Roguecraft DX Steam Review
We’ve been reporting on Roguecraft DX for over a year now, first reporting on the isometric 3D roguelike’s release on Evercade. Then, earlier this year, we reported that the game...
Squint
Drag a box on your screen and ask AI about it Discussion | Link
Bittensor’s Real Experiment Is Paying Markets to Produce Intelligence
Bittensor is easier to understand as a market for measurable digital work. Here’s how subnets, validator scoring, alpha tokens, and TAO incentives fit together.
Why research personas despite RL scaling?
RL seeming to shape much of the motivations and behaviour of the agents, "washing out" their initial personas — c.f. Thoughts on the persona selection model (Sam Marks, 24th Sep...
Mux: Turn Your Video Into Context
My thanks to Mux for sponsoring last week at DF. Video isn’t just something to stream; it’s structured data you build with. Mux Robots — hosted AI workflows — turns...
Welcome to Postreads
Discover and follow the best content from across the web, all in one place. Create an account to start building your personalized feed today and never miss out on great reads.
Tiny Generals
Asynchronous hex-based strategy. Command armies, capture bases, and outthink opponents at your own pace.
Play NowSupport Postreads
Enjoying the service? Help me keep it running and improve it further by buying me a coffee!
Buy me a coffeeContent Timeline
114,983 articles since 2008
Trending Now
Top 5 clicked items this week out of 1,637 added this week
Freshly added
New feeds to discover
